AI News

Weakening Neurons: An Input-Output Functionality in Transformers with Outsize Influence (arxiv.org)

CoRe-MARL: Cooperative Redistribution Under Unknown Dynamics Using Recurrent Multi-Agent Reinforcement Learning (arxiv.org)

Rethinking Critic Learning in PPO: Understanding and Mitigating Value Flattening (arxiv.org)

A Convergence Framework for Deep $V$-Learning: Error Propagation and Sharp Action-Gap Bounds (arxiv.org)

WaveTLM: Reliable Time-Series Language Modeling through Task Compilation (arxiv.org)

Interpretable Multi-Instance Learning Enables Early Prediction of Key Molecular Alterations from Routine Flow Cytometry in Acute Myeloid Leukemia (arxiv.org)

Physics-based prediction, uncertainty quantification and decision-making for IN718 crystallographic texture intensity across LPBF defocus regimes (arxiv.org)

Preventing Model Collapse: A Fisher-Rao Perspective on the Dynamics of Training with Synthetic Data (arxiv.org)

Changepoint-Aware World Models: Detecting Dynamics Shifts and Recovering by Forgetting Stale Replay in Model-Based RL (arxiv.org)

FedGuide: Diffusion Prior Alignment and Value Baseline Guidance for Heterogeneous Federated Reinforcement Learning (arxiv.org)

The Automaton Underneath: The Additive Input Pathway Is a Parasitic Attractor for State Tracking in Householder Linear RNN (arxiv.org)

One Axis, No Brake: Self-Knowledge Limits the Filtering of Harmful Peer Conformity in LLMs (arxiv.org)

Capability Emergence Can Be Forecast: Per-Seed, In Advance, With Calibrated Intervals, Certified False Alarms, and a Blind Pre-Registered Gate (arxiv.org)

CompileRover: Revolutionizing Virtual Machine Compiler Optimization with a Tri-Role LLM-Driven Framework (arxiv.org)

TwinMark: A Unified Watermark for Provable Survival Under Feature and Logit Distillation (arxiv.org)

Integrated Optimization of Automated Warehouse Operations and Last-Mile Transport for Differentiated On-Demand Delivery (arxiv.org)

LightSleepX: A Lightweight, Inception-Based Dual-Modal Network for Sleep Staging (arxiv.org)

Double descent is the principle of least action (arxiv.org)

Evidence-Grounded Agentic Formulation Development in an Autonomous Laboratory (arxiv.org)

How Model Growth, Recursion, and Boundary Operators Influence Scaling Exponents (arxiv.org)

A Calibrated Instrument for Measuring How Inference Optimizations Affect Output Quality (arxiv.org)

QEMScore: How Much Does the Measurement Add to Learned Quantum Error Mitigation? (arxiv.org)

Zing-0.5: Toward Playable Worlds with Real-Time Joint Action and Text Control (arxiv.org)

TACTICS: Taxonomy-Aware Intelligent Corpus Sampling for Machine Translation (arxiv.org)

Mixed-Integer Nonlinear Differentiable Predictive Control for Underground Pumped Hydro Energy Storage Systems (arxiv.org)

ASPIRE: Asynchronous Batched Self-Speculative Decoding for Long-Context LLM Inference (arxiv.org)

Does AI Assistance Leave a Temporal Fingerprint? Detecting Overreliance in AI-Assisted Writing and Programming (arxiv.org)

EDCT-Bench: Uncovering Faithfulness Gaps in VLMs via Explanation-Driven Counterfactual Testing (arxiv.org)

Whom Do AI Agents Work For? Role Assignment Induces Sponsorship Bias in LLM Recommenders (arxiv.org)

Newer Is Not Fairer: Gender Stereotyping in Text-to-Image AI Across Model Generations (arxiv.org)

FuseFi: Combining Irregularly Sampled CSI from Diverse Communication Packets and Frequency Bands for Wi-Fi Sensing (arxiv.org)

Scalable Music Cover Retrieval Using Lyrics-Aligned Audio Embeddings (arxiv.org)

Persistent Magnitude Homology for Quantitative Equational Theories (arxiv.org)

Which Histories Matter for Time Series Forecasting? Learning Predictive Relevance with Future Supervision (arxiv.org)

On the Identifiability of Mixed Ordinal and Exponential Family Causal DAGs under Linear Parametric Models (arxiv.org)

Symmetry without a manifold: intrinsic dimension on orbits (arxiv.org)

Attention Dispersion as a Diagnostic Signal for Hallucination in Large Language Models (arxiv.org)

RecMorph: Topology-Guided Spatial Recurrence for Generalized Morphology Control (arxiv.org)

HiLNO: A Hierarchical Latent Neural Operator with Multi-Scale Supervision for PDEs on General Geometries (arxiv.org)

TabICLv2: A better, faster, scalable, and open tabular foundation model (arxiv.org)

Patch the Distribution Mismatch: RL Rewriting Agent for Stable Off-Policy SFT (arxiv.org)

TabPFN-3.5: Technical Report (arxiv.org)

Modular Deep Learning Mechanisms for Auditable Next-Day Wildfire Spread Prediction (arxiv.org)

Approximating Measures on Function Spaces: Transport and Truncation (arxiv.org)

Random tilts to find stationary points in stochastic convex optimization (arxiv.org)

Efficient Robust Learning at the Information-Theoretic Limit (arxiv.org)

Fenchel-Young Duality Gaps: Certified Early Stopping for Regularized Inverse Problems (arxiv.org)

Pay Only for Disagreement: Certified No-Regression Verdicts for Model Updates with Matching Label-Complexity Bounds (arxiv.org)

A General Kernel Framework for Non-CND Distance Measures Using |D|-Dimensional Sparse Landmark Embeddings (arxiv.org)

Echo: Learning-based Matching Decompilation using Trusted Back Translation (arxiv.org)